Efficiency of Bartlett and Levenes Tests for Testing Homogeneity of Variance Under Varying Number of Replicates and Groups in One – Way ANOVA
B. Odoi1, S. Samita2, S. Al-Hassan3, S. Twumasi-Ankrah4

1B. Odoi, University of Mines and Technology, Tarkwa, Ghana.

2S. Samita, University of Peradeniya, Sri Lanka.

3S. Al-Hassan, University of  Mines and Technology, Tarkwa, Ghana.

4S. Twumasi-Ankrah, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana.

Manuscript received on 13 April 2019 | Revised Manuscript received on 20 April 2019 | Manuscript Published on 26 July 2019 | PP: 1219-1223 | Volume-8 Issue-6S4 April 2019 | Retrieval Number: F12500486S419/19©BEIESP | DOI: 10.35940/ijitee.F1250.0486S419

Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: In numerous investigates, regardless of whether or not beneficial or hypothetical, an inquiry that emerges in performing examination of change is the Homogeneity of Variance suspicion. Severa analysts have checked out terrific assessments for trying out homogeneity of fluctuation and reached determinations dependent on the power and shortcomings of these checks with out considering the effectiveness of these exams below transferring reproduces and gatherings. This paper is devoted to looking on the productivity of Bartlett and Levenes assessments for trying out homogeneity of change depending on each power and kind I blunder beneath transferring variety of reproduces and gatherings (with same contrasts organized with the aid of strategies) for single course ANOVA through one thousand reproductions of arbitrary ordinarily disseminated facts for each mixture of imitates and gatherings. The aftereffects of the exam uncovered that, all in all, type I blunder price diminishes from zero.1 to zero.05 and after that remaining elements quite tons in a degree as the amount of repeats builds unbiased of the quantity of gatherings. Anyways, it seems while range of gatherings is three or a good deal less, Levenes take a look at is better, and Bartlett take a look at is higher when variety of collecting increments. As for kind II blunder, each check brought approximately consistent abatement in mistake rate as sort of recreates and gatherings increments. Be that as it may, each check executed inadequately even with 30 duplicates for every collecting irrespective of the truth that Bartlett test achieved advanced to Levenes at better variety repeats. Turning into test want to be selected by using structure of the data and want to kind I mistake and type II blunder.

Keywords: Type I Error, Type II Error, Simulation, Testing Assumptions.
Scope of the Article: Applied Mathematics and Mechanics